CN-120765914-B - Mouse trapping system based on YOLO mouse hole detection model

CN120765914BCN 120765914 BCN120765914 BCN 120765914BCN-120765914-B

Abstract

The invention relates to the technical field of image processing, in particular to a mouse trapping system based on a YOLO mouse hole detection model, which is provided with a feature acquisition module, a print analysis module, a mouse hole identification module, a model judgment output module and a mouse trapping regulation and control module, wherein the feature acquisition module marks a suspected mouse hole to perform print identification, the print analysis module screens an associated print group, determining the activity track of a suspected rat hole, determining the activity tendency characterization quantity through a rat hole identification module, screening characteristic marks so as to identify characteristic activity tracks, screening characteristic rat holes, outputting the division result of the rat hole system through a model judgment output module, determining the growth tendency coefficient through a rat trapping regulation module, and regulating the rat trapping frequency of the rat hole system. The method and the device realize the judgment of the activity of the rat holes according to the intermittent imprinting around the identified rat holes, perform joint analysis on the active rat holes, adaptively adjust the rat trapping frequency, and improve the reliability and the flexibility of the rat trapping system.

Inventors

LUO XIAOLING
Xie Tianshuo
HAN YANG
LI JIANRONG
PAN XIN
YU WENYUAN

Assignees

内蒙古农业大学

Dates

Publication Date: 20260508
Application Date: 20250708

Claims (10)

1. A mousing system based on YOLO mousehole detection model, comprising: The characteristic acquisition module is used for marking suspected rat holes according to the surface images of all sub-detection areas on the grassland and performing imprinting recognition in a first preset range set according to the suspected rat holes as a reference; the imprinting analysis module is connected with the characteristic acquisition module and is used for screening a plurality of correlation imprinting groups of the suspected rat holes according to the identified imprinting and determining the activity track of the suspected rat holes according to the position distribution of the correlation imprinting groups; The rat hole identification module is respectively connected with the characteristic acquisition module and the footprint analysis module and is used for determining an activity tendency characterization quantity according to the footprint distribution condition of the activity track, screening out characteristic footprints, identifying a characteristic activity track based on the activity tendency characterization quantity and the characteristic footprints and screening out characteristic rat holes according to the characteristic activity track and the soil characterization quantity of the suspected rat holes; The model judgment output module is connected with the rat hole identification module and is used for determining the division result of the rat hole system according to the position distribution of each characteristic rat hole; The mousing regulation and control module is respectively connected with the characteristic acquisition module and the model judgment output module and is used for determining a growth tendency coefficient according to the soil characterization quantity of the characteristic rat holes in each rat hole system, adjusting the mousing frequency of the rat hole system based on the growth tendency coefficient and pushing the mousing frequency to an output end.
2. The YOLO mousing system based on the mouse hole detection model according to claim 1, wherein the print analysis module is used to screen a plurality of associated print groups, wherein, The print analysis module screens two adjacent prints as an associated print group based on a judging result that the interval distance between the two adjacent prints meets the condition of the associated print group; the condition of the associated print group is that the interval distance between two adjacent print groups does not exceed a preset first interval distance threshold.
3. The YOLO rat hole detection model based rat trap system of claim 2, wherein the footprint analysis module is configured to determine an activity trajectory of the suspected rat hole, wherein, The moving track is determined according to a plurality of head-to-tail connected associated print vectors, the associated print vectors are constructed by taking the print with the maximum space between the two adjacent prints and the suspected rat hole as a vector starting point and the print with the minimum space between the two adjacent prints and the suspected rat hole as a vector ending point.
4. A mouse trapping system based on a YOLO mouse hole detection model as claimed in claim 3, wherein the mouse hole recognition module is used for determining activity tendency characterization quantity and characteristic print, wherein, The rat hole recognition module is used for calculating the interval distance between each print in the moving track and the suspected rat hole, determining the maximum value of the interval distance as the moving tendency characterization quantity of the moving track, and determining the print with the minimum value of the interval distance as the characteristic print.
5. The YOLO mousing system based on the mouse hole detection model according to claim 4, wherein the mouse hole recognition module is used to recognize a characteristic moving trace, wherein, The rat hole recognition module is used for recognizing the activity track as a characteristic activity track based on the activity tendency characterization quantity of the activity track and the judging result that the characteristic trace accords with the characteristic activity track condition; the characteristic activity track condition is that the activity tendency characterization quantity exceeds a preset activity tendency characterization threshold, and the interval distance between the characteristic print and the suspected rat hole does not exceed a preset second interval distance threshold.
6. The YOLO rat hole detection model based rat trap system of claim 5, wherein the rat hole identification module is configured to determine a soil characterization of a suspected rat hole, wherein, The rat hole identification module is used for determining gray values of all monitoring points in the sub-detection area where the suspected rat hole is located and gray values in a second preset range based on the surface image, determining the absolute value of the difference between the minimum value of the gray values in the second preset range and the average value of the gray values in the sub-detection area as the soil characterization quantity of the suspected rat hole, and determining the second preset range based on the suspected rat hole.
7. The YOLO rat hole detection model based rat hole detection system of claim 6, wherein the rat hole identification module is configured to screen characteristic rat holes, wherein, The rat hole recognition module screens the suspected rat holes as characteristic rat holes based on the judging result that the characteristic activity track and the soil characterization quantity of the suspected rat holes accord with the characteristic rat hole conditions; The characteristic rat hole condition is that the number of characteristic moving tracks exceeds a preset number reference value, and the soil characterization quantity exceeds a preset soil characterization reference value.
8. The YOLO rat hole detection model based rat hole trapping system of claim 7, wherein the model determination output module is configured to divide each characteristic rat hole into different rat hole systems, wherein, The model judging and outputting module calculates the interval distance between any characteristic rat hole and other characteristic rat holes, and divides the characteristic rat holes with the interval distance not exceeding a preset third interval distance threshold into the same rat hole system, wherein each characteristic rat hole only exists in a unique rat hole system.
9. The YOLO mousing system based on the mousehole detection model according to claim 8, wherein the mousehole control module is for determining a growth tendency coefficient, wherein, The mousing regulation module screens the characteristic rat holes as new characteristic rat holes based on the judging result that the soil characteristic quantity of the characteristic rat holes in the rat hole system accords with the new characteristic rat hole conditions; The condition of the newly-generated characteristic rat hole is that the soil characterization quantity exceeds a preset soil characterization threshold value; the growth tendency coefficient is the ratio of the number of newly generated characteristic rat holes to the number of characteristic rat holes in the rat hole system.
10. The YOLO mousing system based on the mousehole detection model of claim 9, wherein the mousehole regulation module is configured to adjust a mousehole frequency of the mousehole system, the mousehole frequency being in positive correlation with the growth tendency coefficient.

Description

Mouse trapping system based on YOLO mouse hole detection model Technical Field The invention relates to the technical field of image processing, in particular to a mouse trapping system based on a YOLO mouse hole detection model. Background The grassland rat damage is a global problem threatening ecological safety and sustainable development of animal husbandry, in recent years, the ever-increasing number of rat holes on the grassland is large, the types of the grassland rat damage are mostly Brinell field rats, the holes comprise a plurality of holes, the grassland desertification is aggravated, the yield of grasslands is also influenced, the increase of the number of rats also threatens the life of people, the traditional rat damage monitoring depends on manual ground stepping, the manual inspection consumes a great amount of manpower and material resources, the rat hole identification is influenced by the experience of investigators, the error rate is high, the real-time tracking of the activity of the rats is also impossible, dynamic changes such as rat hole expansion and migration are difficult to monitor, prevention and control lag is caused, along with the development of deep learning technology, a target detection model based on YOLO is gradually applied to rat hole recognition, but factors such as illumination change and vegetation shielding can cause misjudgment, the activity of the rat hole cannot be judged by the recognized rat hole, a core rat hole needing to be prevented and controlled preferentially cannot be recognized accurately, the reliability and flexibility of a rat trapping system are affected, and therefore improvement of the reliability of monitoring rat activities and the flexibility of rat trapping decision regulation are technical problems to be solved urgently. For example, the Chinese patent application publication No. CN117746504A discloses a behavior analysis device and a behavior analysis method for modeling an experimental murine animal skeleton based on the YOLO and ST-GCN algorithm, when a user uses the device in other places such as a raising box of the experimental murine animal, the device is matched with a USB camera module which can be connected with a main board to collect images or videos of the experimental murine animal, the images or videos are sent into a behavior analysis system, the system firstly realizes target detection and posture estimation through the YOLO algorithm model, automatically captures dynamic skeleton points, accurately acquires skeleton point data of the experimental murine animal, then carries out dynamic skeleton modeling on data obtained from the previous model through the ST-GCN algorithm model, obtains behavior data such as behavior recognition, centroid data, limb length, motion trail and the like of at least five action modes such as target modification, feeding and the like, and realizes behavior analysis of the experimental murine animal. The following problems also exist in the prior art: The prior art can not judge the activity of the rat holes according to the intermittent imprinting around the identified rat holes, can not carry out joint analysis on the active rat holes and adaptively adjust the rat trapping frequency, and affects the reliability and the flexibility of the rat trapping system. Disclosure of Invention Therefore, the invention provides a mouse trapping system based on a YOLO mouse hole detection model, which is used for solving the problems that the prior art cannot judge the activity of a mouse hole according to the discontinuous imprinting around the identified mouse hole, cannot perform joint analysis on the active mouse hole, adaptively adjust the mouse trapping frequency and influence the reliability and the flexibility of the mouse trapping system. In order to achieve the above object, the present invention provides a mousing system based on YOLO mousehole detection model, comprising: The characteristic acquisition module is used for marking suspected rat holes according to the surface images of all sub-detection areas on the grassland and performing imprinting recognition in a first preset range set according to the suspected rat holes as a reference; the imprinting analysis module is connected with the characteristic acquisition module and is used for screening a plurality of correlation imprinting groups of the suspected rat holes according to the identified imprinting and determining the activity track of the suspected rat holes according to the position distribution of the correlation imprinting groups; The rat hole identification module is respectively connected with the characteristic acquisition module and the footprint analysis module and is used for determining an activity tendency characterization quantity according to the footprint distribution condition of the activity track, screening out characteristic footprints, identifying a characteristic activity track based on the activity tendency chara